RECENT TOPICS » View all
I'm getting into using sub2srs to study from dorama very soon. Would it work to train correct output if what I did was look at a sentence and read it, then close my eyes and be able to reproduce it from memory out loud. This can come from any native material so long as it's not from a sentence pack per se.
I've found a lot of the time if I look at a sentence then close my eyes I can't spit it back out again but it only takes a little bit of drilling right then and there to forge the pathway in my brain to be able to do so. So if I do this for every sentence I wanted to be able to say or whatnot I guess they'd slowly begin to form a useable base for output?
IceCream wrote:
sure. but why not just actually learn the words instead instead of spending the time on the boring sentences?
Because it would deprive the words of their context (not all words require context, but many do) and because it wouldn't help you practice producing and interpreting Japanese sentence structure. I also insist on using voice clips read by a native speaker for every sentence when possible (smart.fm is great for this), so it also helps me practice listening and pronunciation of complete sentences, not isolated words.
magamo wrote:
[...] you need to learn language register, context, and other subtleties through exposure to native material. As you just pointed out in your post with 熱い水, it's the contexts, collocations, and other subtleties that are really important to sound like a native speaker.
Apparently the 10,000 sentences Katzumoto is talking about are all from real native material, so they're richer in context and whatnot while pre-made sentence packs are very poor in this regard. As Nukemarine said, each pack is for a purpose, so they should be better in one regard. But they're not that great when it comes to the essential to achieve native fluency.
Context can only do so much for you, though. For example, if you see a sentence with the word お湯 -- whether it's from a sentence pack or real natve materials -- it's still not going to tell you that you can't say 熱い水. Especially in native materials, there won't be a footnote or anything to tell you that you have to say お湯 instead of 熱い水. The only way to learn this to either be extremely observant (more than I would be) or, more likely, simply be told that people just don't say 熱い水.
And that's the essential problem here: whether you're learning from sentence packs or not, it's not always easy to notice the most important or interesting things about the sentence.
There's also the point that when you put a sentence in your SRS from native materials, you'll often deprive it of its original context, so at that point it might as well have come from a sentence pack anyway.
Note that I'm not saying it's not necessary to use native materials to learn Japanese! You can never master any language without observing how it's actually used (and even if you could, what would be the point?). I'm just saying that I don't think that that has much to do with the "10,000 sentences in an SRS" thing.
I think the best way to practice production of Japanese, or any other language, is to get a journal at lang-8.com and write in it. Then, if you make mistakes, native speakers will correct you, so if you end up saying 熱い水, somebody will replace it with お湯 for you, and then you'll know that you can't say 熱い水.
- Kef
Last edited by furrykef (2009 November 02, 4:47 am)
the smartfm / KO sentences don't really help in practising or producing natural sentence structures to a great extent in any case, i think. (i mean, for things you might actually want to say). If you're using the sentence packs to learn the words so that you can recognise them in real contexts, then just doing word recognition, and using the extra time you saved to enter sentences from real contexts into your deck is far more beneficial imo.
I dunno, I don't find the smart.fm sentences to be excessively artificial. Also, the thing here is that many words are only used in certain context. Imagine trying to learn how to use the English word "shrouded" (as in "shrouded in darkness") without context, instead learning it as something like "obscured, hidden". Chances are, you'd use it wrong, because you wouldn't have any exemplary usage to learn from.
For pitch, single words are often easier to practise with than whole sentences.
I wouldn't say that, considering that there are plenty of words that are pronounced identically in isolation but not when followed by another word (that is, words that are accented on the final mora).
- Kef
Last edited by furrykef (2009 November 02, 5:26 am)
@furrykef
I think you're missing the point a bit. I'm not saying native stuff can teach you everything in one shot. You should refine your understanding through exposure.
I think 10,000 only means that if you keep picking up choicer sentences from context-rich material while honing your intuition by reading/listening to tons of native stuff, you'll have been fluent by the time you have SRSed that amount of sentences. Since sentence packs lack a certain kind of information, you can't become fluent by SRSing 10,000 pre-made sentences.
That's why I think it's not odd or strange not to count pre-made sentences for the 10,000 sentences. In the context of sentence mining, the number of sentences you have SRSed is supposed to roughly indicate how much you have consumed native material.
Last edited by magamo (2009 November 02, 5:44 am)
Magamo hit on my point pretty well. I'm not saying these pre-generated list are not useful. They've been pretty damn useful for me. However, it was during my subs2srs phase that it became blatantly obvious that isolated sentences, while great to get across grammar and vocabulary, are horrible when it comes to instilling the ability to fluidly speak and listen.
Think about it this way, I used "宇宙人は存在すると思いますか?" to get across 存在, and I then looked up 宇宙 in another sentence which mentioned 謎 so I looked that up in another sentence in Core 6000 (recursive studying).
However, I still don't know what was asked or responded to any of these sentences. Though with the sentence "ムキなんなよ。たかが暇潰しでしょ。" I know was preceded by 道明寺 not wanting to stop, and あきら making fun of him not handling a little school. If I need to know むき or 暇 or 潰し with a good sample sentence, then there's my vocabulary pack. But listening to an hour of 花より団子 after dissecting it, versus an hour of iKnow sentences is not close relation at all with what helps conversation.
So, while I might know 3000 words thanks to the 3000 sentences, I'm not going to say I'm 3000 sentences into the sentence method. Right now, out of four TV shows, I'm at about 1000 sentences and dropping (I still keep deleting them over time). But in the 5 months time of getting those thousand sentences I've advanced more than what happened in 2700 "sentences" from Tae Kim and iKnow.
Still those 2700 sentences did prepare me for what was to come, make no mistake. So I encourage "packs", but I'll say you'd be better not counting anymore than you'd count the 2000 kanji from RTK. And at a certain point, it's best to use packs as a resource to be called upon only when you need a part of it, utilizing recursive studying.
By the way, this is all based on my personal experience. I'm still horrible at Japanese, so take with a large dose of salt.

